Desiderata #6: links and commentary
Kurzweil returns, Shannon's "ultimate machine," Amis on nuclear war
1. This last month The Intrinsic Perspective published:
How AI’s critics end up serving the AIs: On the obscuring effects of AI criticism
When social media controls the nuclear codes: An internal dialogue
2. Why do humans like stuffed animals so much? Perhaps it’s because their softness and size triggers in us the same feeling holding a baby does. This is backed up by recent research showing that the trigger for maternal affection in macaques is a soft object of the right size—hard objects don’t trigger such feelings.
It’s interesting research. At the same time, the thought of mother macaques in cages tightly gripping their stuffed dolls affects me. I once spent a summer at NYU running neuroscientific experiments on macaques, and the whole thing unsettled me deeply, and that experience is why I don’t do animal research.
3. The AI-ification of everything continues.
Not only that, but in the future famous actors might return again and again to the screen. E.g., Bruce Willis has already done ads where he’s just an AI:
Deepcake’s spokesperson explains that their company created his digital twin for 2021 ad campaigns. . .
“I liked the precision of my character,” Willis said, according to the quotes on the Deepcake site. “It’s a great opportunity for me to go back in time. The neural network was trained on content of Die Hard and Fifth Element, so my character is similar to the images of that time.”
In March, Willis’ family posted a statement to social media to announce that the star would be “stepping away” from his acting career after being diagnosed with aphasia, which had impacted his cognitive abilities.
I don’t think this is good for culture. Actors, writers, artists, should all have their time, and then exit the stage. This is a recipe for cultural stagnation—imagine in 2050 an action movie blockbuster starring an AI Bruce Willis. Die Hard VII. Ugh.
4. And in more hints of cultural stagnation, congress is becoming a gerontocracy at a parabolic rate:
5. The winners of the $20,000 EA movement criticism contest, which “Why I am not an effective altruist” was entered into, have been announced. I did not win, but that was expected. As I said in the original post:
Unfortunately, my criticism is how the utilitarian core of the movement is rotten, and I think that’s the part anyone in the movement least wants to see criticized, since there’s almost nothing they can do about it if the core is indeed wrong.
I liked a lot of the winning entries that I read. However, my meta-criticism of the contest is that if you just handed someone the texts of the winners without describing the contest to them, it would be difficult to guess this was “Awards for EA criticism” vs. “Awards for independent EA work”—it’s all extremely in-line with the methods and assumptions of standard EA, and a lot boiled down to “You could be EA-ing even harder.”
6. In terms of actually substantial criticism of EA, Maximum Progress has an interesting argument that showcases a new failure in utilitarianism if we happen to discover aliens. If we find alien life is quite common, this would mean that from a strict utilitarian perspective protecting humanity from existential risks isn’t much of a concern, since if we died some other civilization would likely eventually take our place (if not on Earth than in our surrounding galactic neighborhood). This shows off one of the key weaknesses of utilitarianism as a moral theory: its inability to justify a preference for the human over the inhuman.
7. Replication shows that universal daycare is not so great for outcomes, with a surprisingly strong effect.
To be perfectly honest, while some conservatives are saying this means that universal daycare is uniquely bad, I don’t think that’s what it means. I’d actually expect to see similar effects if we could compare public school, or even private school, to some sort of ideal version of education—but who is the control group? Religiously homeschooled kids? Our entire system of education is so totalizing and self-similar it’s difficult to see if it even works, since there’s basically no alternative. Although I openly admit my bias: I have a very pessimistic view of the subject. In my memory, entering the daycare → preschool → school pipeline felt a lot like being drafted into the army. You wake up early, you follow orders, and the pay is shit. This pessimistic view of contemporary education is why I originally wrote about aristocratic tutoring (I’m excited to say there will be more upcoming entries on that topic).
8. Books are getting shorter over time, argues The Spectator, with contemporary literature being especially thin—one might even say slight, or perhaps, to match our age of brands, lite.
A brief history of the novel-by-length begins in the 18th century with Daniel Defoe, Henry Fielding and particularly Samuel Richardson giving a lot of bang for your buck; Clarissa comes in at just a shade under a million words. The Victorian versions became less episodic, which cut the length a little, but they remained typically pretty chunky: Great Expectations and Vanity Fair are both just under 200,000 words, Midddlemarch just over 300,000. . .
Then in came modernism – more personal, immediate and urgent – and out went ‘all that David Copperfield kind of crap’, as J.D. Salinger famously put it in The Catcher in the Rye (73,000). For a time books became much shorter and, I’d argue, more readable. A Passage to India is 92,000, Sons and Lovers 91,000, Mrs Dalloway a mere 44,000.
So for most of their history novels were getting progressively shorter. But then came the 1960s, and drugs and free expression. Suddenly the only rule was that there were no rules. . . We live in the ongoing aftermath of that torn-up rule book. Since the 1960s a novel can be so slight that it’s really just a long short story, or it can be a giant: Vikram Seth’s A Suitable Boy is twice as long as Ulysses. . .
So how refreshing it was to see the Booker prize take another turn last month – putting the short in shortlist, as it were. . .